Biggest AI Funding in Europe: Behind the Scenes of Aleph Alpha’s 500M USD Series B
DDVC #60: Where venture capital and data intersect. Every week.
👋 Hi, I’m Andre and welcome to my weekly newsletter, Data-driven VC. Every Thursday I cover hands-on insights into data-driven innovation in venture capital and connect the dots between the latest research, reviews of novel tools and datasets, deep dives into various VC tech stacks, interviews with experts, and the implications for all stakeholders. Follow along to understand how data-driven approaches change the game, why it matters, and what it means for you.
Current subscribers: 14,680, +180 since last week
Brought to you by VESTBERRY - The Portfolio Intelligence Platform for Data-driven VCs
Watch a short video showcasing the automation of monthly portfolio performance reviews through workflows using ChatGPT, Vestberry, Slack, and Gmail via make.com! Discover a new way to stay updated on your portfolio's performance and offer support to your portfolio companies with the power of AI.
We’re incredibly excited to double down and participate in the $ 500M Series B funding round of our portfolio company Aleph Alpha. The oversubscribed round was co-led by Innovation Park AI (IPAI), Robert Bosch Ventures, and Schwarz Group, followed by Hewlett Packard Enterprise (HPE), SAP, Christ&Company, Burda Principal Investments as well as existing investors.
It’s not only the biggest AI funding round ever in Europe, but also a very special setup, given the partners involved. In this post, I’d like to share the full story of how we identified the next generative AI champion with the help of AI, our initial belief and why we invested, what happened within the 2.5 years until today, and most importantly, why we doubled down in such a special funding setup, and what the future holds. Let’s dive in!
Leveraging AI to Find the Next AI Champion
This newsletter is all about the intersection between VC and AI. Following my PhD on “Machine Learning and the Value of Data in Venture Capital”, I conceptualized and productized the first version of what we call “Earlybird’s EagleEye”. Between then and early 2022, however, I was the only person using it.
Why? Because I understood early enough that blockers for adoption of data-driven approaches in the investment world would not be technical but cultural in nature. You need top-down support from the General Partners and only get one chance to change the workflows of your investment team. If you fuck it up, you won’t get another try. Therefore, we developed EagleEye mostly behind the scenes and only rolled it out at a more mature stage than you would typically do. Contrary to the “zero-to-one” approach of fast iterations and short feedback cycles.
While I sourced an increasing number of opportunities during the beta test phase, none made it through our investment committee. None, except for Aleph Alpha. We sourced the company through their Handelsregister (=German public register) registration in 2019 but didn’t prioritize it until the end of 2020 when sufficient online data became available to improve their “likelihood of success” score and highlight it in our system.
Early Conviction: Why We Led the € 23M Series A
I vividly remember my first call with Jonas, the CEO and Co-Founder of Aleph Alpha, in spring 2021. We immediately connected on a deep content level and shared a lot of theses around the future of AI. The follow-up call with his co-founder Samuel perfectly complemented my first impression. From initially an hour scheduled, we ended up discussing and playing around with their models all afternoon.
I was deeply impressed by both of them, their profound expertise, entrepreneurial drive, and unique positioning in a new, potentially unlimited market. However, Generative AI was not a thing back then and the founders’ ask for € 20m+ without any commercial traction didn’t make the discussions within our partnership easier.
After some back and forth, a lot more research and references, tough negotiations, and syndication efforts on all sides, we eventually ended up leading Aleph Alpha’s € 23M Series A round in June 2021. Only 6 months after their first funding round end of 2020. I outlined our initial investment ratio here; some extracts:
Together they had the clear vision to establish a European alternative to OpenAI and BAAI, and establish a globally leading AI-research institution with European values at its core (..)
Aleph Alpha decided to build not only an independent and open multi-language alternative to the closed US and Chinese offerings but add functionality that makes the integration, (ethical) alignment and innovation based on large models easier, more transparent and robust.
Knowing that developers seek more than performance, Aleph Alpha decided to pick a select few light-house customers with a strong pull-dynamic and broadly relevant use cases to gradually educate the market, better understand customer needs and ultimately identify reusable components (..)
Moving closer towards large generalizable models/AGI requires significant computing power/resources, comprehensive training datasets and the ability to overcome unique engineering challenges (=talent).
“These models are so adaptable and flexible and their capabilities have been so correlated with scale we may actually see them providing several billions of dollars worth of value from a single model, so in the next five years, spending a billion in compute to train those could make sense” - Bryan Catanzaro, VP of Allied Deep Learning Research at NVIDIA
In a nutshell, we gained strong conviction that Aleph Alpha can stay competitive on performance by accessing compute, talent, and data, yet leaning more into B2B relevant aspects such as independence, trustworthiness, transparency, and explainability - all by incorporating values and norms of the Western World, and allowing their customers to decide what’s right or wrong. We assumed this to be their edge in the market.
Why this seemed so important? Well, as Jonas tends to say “we want to have a hand on the wheel”. Generative AI will soon be controlling every single product we use. Giving up sovereignty and becoming no more than a customer, we leave the directive to other players based in China or the US, similar to what Europeans did with social media platforms. We assumed that Aleph Alpha’s sovereign and independent positioning would help to gain trust with customers (and yes, by now we know it does).
Essentially, our initial thesis boils down to “becoming the critical infrastructure for customers who run our critical infrastructure”. Said differently, gaining trust and providing competitive generative AI technology to customers across governments and industries such as finance, insurance, legal, healthcare, and industry, among others.
Although all of our initial assumptions seem to still hold 2.5 years later, one of them was partially off: “Entry barriers will increase and generative AI will be centralized.”
Partially off, because we did not expect open-source variants such as Llama to drive democratization, eventually leading to convergence of LLM performance. Today, we know that LLMs will become a commodity very soon. It’s comparably cheap and simple to train them.
Partially right, because independent of the democratization aspect, the value created seems to stay centralized with a “few take all logic”, just as for the hyper scalers. Looking at the actual value captured in $$, OpenAI is indisputably the giant in the room and it becomes increasingly more difficult to compete with them.